Code Changes - Updated the embedding models

app.py

@@ -29,6 +29,7 @@ from utils import (
 
 from constants import (
     CHUNKING_STRATEGIES,
+    EMBEDDING_MODELS,
 )
 
 from ragbench import RAGSystem, RAGEvaluator
@@ -79,8 +80,6 @@ for item in ragbench_details.keys():
 
 available_subsets = list(ragbench_details.keys())
 
-# chunking_strategies = ["SentenceBasedLangchain", "Hybrid"]
-embedding_models = ["BAAI/bge-large-en-v1.5", "intfloat/e5-large-v2"]
 generator_models = ["mistralai/Mistral-7B-Instruct-v0.2"]
 evaluators = ["llama"]
 
@@ -314,7 +313,7 @@ with gr.Blocks(
         with gr.Column(scale=3):
             gr.Markdown("### ⚙️ Chunking and Model Selection")
             chunking_dropdown = gr.Dropdown(choices=CHUNKING_STRATEGIES, label="📦 Chunking Strategy", value="SentenceBasedLangchain")
-            embed_dropdown = gr.Dropdown(choices=embedding_models, label="🔗 Embedding Model", value="BAAI/bge-large-en-v1.5")
+            embed_dropdown = gr.Dropdown(choices=EMBEDDING_MODELS, label="🔗 Embedding Model", value="BAAI/bge-large-en-v1.5")
             retriever_dropdown = gr.Dropdown(choices=generator_models, label="🧭 Generator Model", value="mistralai/Mistral-7B-Instruct-v0.2")
 
         with gr.Column(scale=4):

constants.py

@@ -25,6 +25,13 @@ CHUNKING_STRATEGIES = [
     # HYBRID
 ]
 
+
+EMBEDDING_MODELS = [
+    "BAAI/bge-large-en-v1.5",
+    "intfloat/e5-large-v2",
+    "all-MiniLM-L6-v2"
+]
+
 default_json = {
     "relevance_explanation": "Documents 1 and 4 contain useful information for answering the question. Document 1 mentions 'Signal Information' and 'Self Diagnosis', which are relevant to finding signal information. Document 4 provides additional context about signal information, such as it being only available for digital channels.",
     "all_relevant_sentence_keys": [